AnchorFlow：基于潜变量锚点对齐流的免训练三维编辑
AnchorFlow: Training-Free 3D Editing via Latent Anchor-Aligned Flows
把扩散采样中随 timestep 变化的噪声视为潜变量锚点不一致的根源，用源/目标轨迹共享的全局潜变量锚点加松弛对齐损失来稳定编辑方向；做 3D 生成或编辑时可直接借用这一「锚点对齐更新规则」来抑制采样漂移，且无需掩码监督即可保持几何保真。
原文依据 AnchorFlow establishes a global latent anchor shared between the source and target trajectories
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论免训练三维编辑旨在无需模型微调的情况下，依据人类指令修改三维形状，在三维内容创作中起着关键作用。然而，现有方法往往难以产生显著且几何稳定的编辑，这在很大程度上源于扩散采样过程中随 timestep 变化的噪声所引入的不一致潜变量锚点。为解决这些局限，我们提出 AnchorFlow，它建立在潜变量锚点一致性原则之上。具体而言，AnchorFlow 在源轨迹与目标轨迹之间建立一个共享的全局潜变量锚点，并通过松弛的锚点对齐损失与锚点对齐更新规则来施加一致性约束。该设计确保变换在整个编辑过程中保持稳定且语义忠实。通过稳定潜变量参考空间，AnchorFlow 能够实现更为显著的语义修改。此外，AnchorFlow 无需掩码。在没有掩码监督的情况下，它仍能有效保持几何保真度。在 Eval3DEdit 基准上的实验表明，AnchorFlow 在多种编辑类型上均能稳定地给出语义对齐且结构稳健的编辑结果。项目主页：https://zhenglinzhou.github.io/AnchorFlow/。
英文原文摘要 1232 字符
Training-free 3D editing aims to modify 3D shapes based on human instructions without model finetuning. It plays a crucial role in 3D content creation. However, existing approaches often struggle to produce strong or geometrically stable edits, largely due to inconsistent latent anchors introduced by timestep-dependent noise during diffusion sampling. To address these limitations, we introduce AnchorFlow, which is built upon the principle of latent anchor consistency. Specifically, AnchorFlow establishes a global latent anchor shared between the source and target trajectories, and enforces coherence using a relaxed anchor-alignment loss together with an anchor-aligned update rule. This design ensures that transformations remain stable and semantically faithful throughout the editing process. By stabilizing the latent reference space, AnchorFlow enables more pronounced semantic modifications. Moreover, AnchorFlow is mask-free. Without mask supervision, it effectively preserves geometric fidelity. Experiments on the Eval3DEdit benchmark show that AnchorFlow consistently delivers semantically aligned and structurally robust edits across diverse editing types. Project page: https://zhenglinzhou.github.io/AnchorFlow/.